The cell walls of gram-negative bacteria contain cross-linked moieties known as peptidoglycans. A number of gram-negative bacteria produce proteins which are covalently linked to the peptidoglycans. Such a protein is referred to as a peptidoglycan-associated lipoprotein (PAL). PALs are present as part of the tol locus in a number of gram-negative bacteria, including Legionella pneumophila (Bibliography entry 1), Escherichia coli (2), Haemophilus ducreyi (3), Campylobacter jejuni (4), Pseudomonas putida (5), Brucella abortus (6), Pseudomonas aeruginosa, Klebsiella aerogenes, Serratia marcescens, Proteus vulgaris, Salmonella typhimurium (7) Actinobacillus pleuropneumoniae (8), Helicobacter pylori (9) Chlamydia pneumoniae (10), and Chlamydia trachomatis (11).
Other PAL-containing bacteria are the Haemophilus influenzae (H. influenzae) bacteria. The H. influenzae bacteria are divided into two groups. Those strains which possess a known capsule are typed by the serological reaction of the capsule with reference antisera. Types a-f have been identified. Strains which fail to react with any of the reference antisera are known as nontypable.
H. influenzae type b (Hib) is the most frequent cause of neonatal meningitis and other invasive infections in the United States (12). The major incidence of childhood meningitis occurs between the ages of one and five years. Sixty percent of those meningitis cases due to Hib occur in children under the age of two (12).
Nontypable H. influenzae (NTHi) is a gram-negative organism which causes a number of diseases, including pneumonia, bacteremia, meningitis, postpartum sepsis, and acute febrile tracheobronchitis in adults (13). NTHi has been reported to cause between 20 and 40 percent of all cases of otitis media seen in young children (14, 15, 16). Children may experience multiple infections due to the same organism since infection confers no long lasting immunity. Current therapy for chronic or repeated occurrences of otitis media includes administration of antibiotics and insertion of tubes to drain the inner ear. NTHi strains have also been implicated as a primary cause of sinusitis (17). Additionally, NTHi causes neonatal sepsis.
Current capsular-based antigenic compositions are ineffective against NTRi. The surface of these bacteria has been shown to be extremely antigenically variable, with the major outer membrane proteins, P1 and P2, being particularly diverse (18, 19). In humans, the presence of serum bactericidal antibodies has been reported to correlate with protection from otitis media caused by sensitive NTHi strains (20).
Candidates for inclusion in antigenic compositions against NTHi should be highly conserved at the amino acid level, surface exposed (in particular, outer membrane proteins), elicit bactericidal antibodies, and be present in all isolates. Previous research has shown that the P6 (also known as PBOMP-1 and HiPAL) (21) protein of NTHi meets all of these criteria. The purified native proteins have been shown to elicit bactericidal antibodies (22, 23, 24, 25) and are conserved antigenically (22, 23, 26, 27).
Evaluation of the genetic sequence of the P6 gene has shown that it is highly conserved among otic NTHi isolates and thus the protein sequence is also highly conserved. Native P6 is a lipoprotein, more specifically a PAL, which is modified at the amino-terminal cysteine with lipids. This protein is present in H. influenzae in relatively small amounts (less than 1% of total outer membrane proteins), making purification from the native organism of useful quantities quite difficult. Thus, a recombinant version of P6 is required for further development as a component in antigenic compositions.
Several laboratories were unable to express lipidated rP6 in large quantities in E. coli (28, 29). As a result, initial recombinant constructs expressing P6 in E. coli could express only a nonlipidated version of the protein. These groups reported that, while the lipidated P6 protein purified from H. influenzae was more immunogenic than nonlipidated rP6 purified from E. coli, it was difficult to engineer a DNA vector which would express lipidated P6 (28, 29); i.e., not better than the low levels of native P6 expressed by H. influenzae. 
Previous attempts to express lipidated rP6 relied on promoters which were not under tight transcriptional regulation, such as trc, taq, lac and PL-C1857. It was theorized that this somewhat leaky transcription led to subtle effects on the E. coli which contributed to low levels of expression of the lipidated protein. Experimental evidence indicated that the ability of signal peptidase II to add lipid to the N-terminus of the protein was not responsible for the low yield of processed P6 (data not shown).
While the P6 protein of H. influenzae has been a primary candidate for inclusion in antigenic compositions against Haemophilus disease (20, 23, 24, 25, 30, 31), the relatively small amounts available from H. influenzae have made recombinant expression of this protein essential. Previous efforts to express lipidated P6 protein in meaningful quantities have been unsuccessful (28, 29). Thus, researchers have focused on the expression and purification of multiple forms of nonlipidated P6.
The antibody response engendered by the nonlipidated rP6 was biologically functional, capable of protecting infant rats from meningitis (28) and eliciting bactericidal antibodies (28, 29), but of a lower magnitude than those elicited by lipidated native P6 (28).
Therefore, there is a need to construct host cell-expression vector systems which express lipidated PALs of gram-negative bacteria. In particular, there is a need to construct host-cell expression vector systems which express lipidated rP6, which can then be included in antigenic compositions against H. influenzae. 